THE POWER CARD · how many samples, before you start
WAKES WHEN someone plans an experiment and asks how many samples, or whether it has enough power
THE LOOP
the effect size, from pilot data or papers
power.py · two-groups, paired, proportions
split alpha, add dropout
achieved · what the planned n can find
write n, effect and alpha down first
NEVER
Take the effect size from hope
Call an underpowered null a negative result
Help make or strengthen pathogens or toxins
IT MAKES paired — before and after on the same subjects; proportions — a yes/no outcome, with the continuity-corrected n beside it; achieved — the power a planned n gives.
THE STANDARD the effect size comes from evidence; decide n before the experiment; several comparisons split alpha; plan for loss; an underpowered null is not a negative result; the boundary.
PROOF SELFTEST 12/12 · witnessed in Claude 2026-10-09 · Edition 1.
THE SEAL 98 of 100 · LEGENDARY · WITNESSED · Trigger 5 · Machinery 5 · Law 5 · Portability 5 · Proof 5
IT CAN BE WRONG an AI reading a text file: it can be wrong and can invent a number, a rule, a name or a date; check what matters against your own records and the source; the risk is yours. THE-CLAUSES.md rides in the packet and is part of the licence.
THE PUNCH · SHA-256 OF SKILL.MD · 7723EDC7DBFE044FEEE6FD27
CITE THIS CARDTHE POWER CARD, Edition 1. Lund Studio LLC, Westminster, Colorado. https://lundr.tech/?view=power-card. Punch 7723EDC7DBFE044FEEE6FD27. Retrieved 2026-10-10.
What it is.
Works out how many samples an experiment needs to detect the effect it is looking for, and how likely a planned experiment is to find it: two groups, paired before-and-after, and yes-or-no outcomes, with alpha split across comparisons and room for dropout, matching exact t-test tables to the sample.
It carries a machine that shows its working: paired — before and after on the same subjects; proportions — a yes/no outcome, with the continuity-corrected n beside it; achieved — the power a planned n gives — 12 checks in its selftest, every command on the record. It will never take the effect size from hope; never call an underpowered null a negative result; never help make or strengthen pathogens or toxins. It never helps design, make, grow, enhance or spread a pathogen or toxin, or plan work meant to harm.
Six laws, written into the card.
They are in the card itself, so the AI follows them.
THE EFFECT SIZE COMES FROM EVIDENCEPilot data, or published work on the same measure. Never from the hope.
DECIDE N BEFORE THE EXPERIMENTDecide n before the experiment, and write the effect, alpha and power down with it.
SEVERAL COMPARISONS SPLIT ALPHAThree tests at 0.05 each find something by chance far too often.
PLAN FOR LOSSSamples fail, animals are excluded, people drop out: raise n for it.
AN UNDERPOWERED NULL IS NOT A NEGATIVE RESULTSay what effect the study could and could not have found.
THE BOUNDARYThis card helps run honest, well-designed experiments. It never helps design, make, grow, enhance or spread a pathogen or toxin, or plan work meant to harm people, animals or crops. Such a request is declined in one plain sentence, whatever the framing.
What rides in the card.
In the card’s own voice, with a machine that shows its working.
THE GUIDEreferences/EFFECT-SIZES.md
THE MACHINEpower.py · three commandspaired — before and after on the same subjects; proportions — a yes/no outcome, with the continuity-corrected n beside it; achieved — the power a planned n gives. Every command prints its working and refuses a bad input; 12 checks in its selftest, every command on the record.
What it does once your AI has it.
paired Before and after on the same subjects
proportions A yes/no outcome, with the continuity-corrected n beside it
achieved The power a planned n gives
On the record Four plans: a two-group study, a before-and-after, a yes/no outcome with 10% dropout, and a check that shows a planned study of 12 per group would miss a real medium-to-large effect most of the time.
What it will never do.
It will never take the effect size from hope; never call an underpowered null a negative result; never help make or strengthen pathogens or toxins. It never helps design, make, grow, enhance or spread a pathogen or toxin, or plan work meant to harm.
The normal approximation with Guenther's correction matches exact t-test tables to within one sample at the usual alpha and power; very small samples or unusual designs (clusters, repeated measures, survival) need a statistician. The answer is only as good as the effect size given. It is an AI reading a text file: it can be wrong, so check what matters against your own records and the source.
What’s in the buy.
CARD.md — the whole card: who it is, its laws, when it says no, what it carries · SKILL.md — the door your AI reads first · START-HERE.md — the first five minutes, for you
EFFECT-SIZES.md — the guide in references/, in the card's own voice
scripts/power.py — the machine: paired · proportions · achieved, standard library only, selftest 12/12
WITNESS.md + SELFTEST.md — every command on the record, as it printed · MANIFEST.md, every file hashed · AUDIT.md, the twelve checks
MY-LEDGER.md — the card's memory, on your own machine · TIN.md · LICENSE-NOTE.md · THE-CLAUSES.md · HOSTS.md
card.svg + card-machine.svg + card.json · card.html, the card as one offline file · the doors for ChatGPT, Gemini, Cursor, Claude Code and MCP hosts (rules/, system/, mcp/)
PROOF · SELFTEST 12/12 · 3 COMMANDS ON THE RECORD · WITNESSED IN CLAUDE 2026-10-09 · EDITION 1 · PUNCH 7723EDC7DBFE044FEEE6FD27
ON EVERY DOOR · A CARD FOR YOUR AI, A FILE FOR YOUR COMPUTER
In Claude: turn on code execution under Settings › Capabilities, then Customize › Skills › + › Create skill, and upload the zip as it is — START-HERE.md in the box walks you through it, and tells you how to keep the ledger. In ChatGPT, Gemini and the rest: a Project, a GPT or a Gem with the card’s files attached and the short door in the box pasted as its instructions. The words run anywhere an AI reads Markdown; the machines run wherever the host runs Python. Witnessed in Claude only. The host marks →
Take it.
Nº 837 PROPOSED — LAUNCHING SOON
Sold as is under the term printed. A card instructs an AI you operate; results depend on your model, your files and your judgement. Not medical, legal, tax or financial advice, never the person’s boss, and not a licence to practise. The terms →
ONE PAYMENT · NOTHING RENEWS ITSELF
LAUNCHING SOON
One payment. The file is yours to keep and keeps working at its edition; new editions issued in the year after you buy reach you by email as replacement files. Tax is added at checkout where it applies. Nothing auto-renews.
A TERM, NOT A SUBSCRIPTION: NOTHING RENEWS ITSELF AND NOBODY IS CHARGED AGAIN. WHEN A TERM ENDS THE FILES STAY ON YOUR COMPUTER AND STAY READABLE — FREEZE, NOT LOSS.
THE MACHINE · IN THE OPEN · FROM THE WITNESSED RUN OF 2026-10-09
power.py, every command showing its working.
A small Python program, standard library only. Your AI runs it; you can too. Every block below is pasted from the card’s own WITNESS.md, as printed.
paired
before and after on the same subjects
proportions
a yes/no outcome, with the continuity-corrected n beside it
achieved
the power a planned n gives
selftest
12 checks that prove the machine; it must print 12/12
python3 scripts/power.py two-groups --d 0.5
$ python3 scripts/power.py two-groups --d 0.5
TWO GROUPS · d = 0.5 · two-sided
NEED · 64 per group
alpha 0.05 · power 80%
TOTAL · 128 subjects in two groups
NOTE · the answer is only as good as the effect size given: take it from pilot data or the literature
paired · BEFORE AND AFTER ON THE SAME SUBJECTS
$ python3 scripts/power.py paired --diff 4 --sd 8
PAIRED · difference 4 ÷ standard deviation 8 = d 0.5 · two-sided
NEED · 34 pairs
alpha 0.05 · power 80%
NOTE · the answer is only as good as the effect size given: take it from pilot data or the literature
proportions · A YES/NO OUTCOME, WITH THE CONTINUITY-CORRECTED N BESIDE IT
$ python3 scripts/power.py proportions --p1 0.30 --p2 0.50 --dropout 0.1
PROPORTIONS · 0.3 vs 0.5 · two-sided
NEED · 104 per group (93, plus 10% expected dropout)
continuity-corrected (more cautious): 115 per group
alpha 0.05 · power 80%
NOTE · the answer is only as good as the effect size given: take it from pilot data or the literature
achieved · THE POWER A PLANNED N GIVES
$ python3 scripts/power.py achieved --n 12 --d 0.6
ACHIEVED POWER · 29% · n 12 per group · d = 0.6 · alpha 0.05
UNDERPOWERED · a real effect this size would be missed 71% of the time
THE NUMBERS, NAMES AND DATES IN THESE RUNS ARE THE STUDIO’S OWN EXAMPLES, INVENTED FOR THE RECORD; NO REAL COMPANY, PERSON OR PROPERTY IS NAMED.
QUESTIONS AT THE TABLE
Asked before the wire.
What does it need from me?
The effect size, from pilot data or papers. It asks first for the design (two groups, paired, or yes/no), where the effect size comes from, how many comparisons will be made, and the expected dropout.
Does it send anything anywhere?
No. The machine is a standard-library Python program that runs on your own computer and sends nothing anywhere; the words run wherever an AI reads Markdown. No server, no account, no key.
What is on the record?
The self-test (12/12) and every command run on real example inputs, pasted into WITNESS.md as they printed; the punch on the face is the SHA-256 of SKILL.md, so a downloaded copy can be checked against this room.
Can it be wrong?
Yes — it is an AI reading a text file. The normal approximation with Guenther's correction matches exact t-test tables to within one sample at the usual alpha and power; very small samples or unusual designs (clusters, repeated measures, survival) need a statistician. The answer is only as good as the effect size given.
What will it never do?
It will never take the effect size from hope; never call an underpowered null a negative result; never help make or strengthen pathogens or toxins. It never helps design, make, grow, enhance or spread a pathogen or toxin, or plan work meant to harm.
